The Lagrangian and the Axion
Introduction
This article writes the axion as the extremum of an action whose interaction term is a total derivative and whose potential is periodic by consistency with the topological charge. It is the action-first companion of The Axion and the Peccei–Quinn Mechanism in Biquaternionic Form, which states the mechanism and the anomalous coupling; this article asks what functional produces the axion's equation, and why the periodicity of the field is compatible with the anomaly.
The equation. The parent's equation is
$$ \Box_\Phi a+\frac{\partial V}{\partial a}=\frac{g_s^2}{32\pi^2 f_a}\sum_{\mu,\nu,a'}F_{a'\mu\nu}F^{a'}_{\mathrm{dual},\mu\nu} , $$
the massive wave equation with a source that is itself a total derivative. The article asks what action has this equation and what the periodicity of $a$ requires of it.
Three things are established.
First, the action. With $a$ the axion field and $V$ its periodic potential,
$$ S[a]=\int d^4r\;\Bigl[\frac12\sum_\mu\frac{\partial a}{\partial R_\mu}\frac{\partial a}{\partial R_\mu}-V(a)+\frac{a}{f_a}\frac{g_s^2}{32\pi^2}\sum_{\mu,\nu,a'}F_{a'\mu\nu}F^{a'}_{\mathrm{dual},\mu\nu}\Bigr], $$
the stationarity condition is the parent's equation, and the interaction term is a total derivative because $F\tilde F$ is one.
Second, the interaction term's total-derivative character is what makes the shift consistent. Under $a\mapsto a+2\pi f_a$ the interaction changes the action by $2\pi$ times the topological charge, which is an integer, so the path integral is unchanged. A coupling that was not a total derivative would make the shift a genuine change of the theory, and the potential's periodicity would be lost.
Third, the potential's periodicity is a consequence and not an assumption. The potential is generated by the same gauge fields the axion couples to, and their topological charge's quantisation forces the potential to be periodic in $a/f_a$ with period $2\pi$. The cosine form is the leading term of the Fourier series the periodicity allows.
What is not new. The axion, the Peccei–Quinn mechanism and the potential are standard. What the corpus contributes is the biquaternionic writing of the periodic central field and the statement of the periodicity's consistency in the algebra's letters.
Conventions
The conventions are restated here so that every sign below is checkable from this page. The algebra is $\mathbb{B}=\mathbb{C}\otimes_{\mathbb{R}}\mathbb{H}$ with $e_t=1$ and $\vec e_\eta^{2}=-e_t$; the axion is a real central scalar, which is the central-scalar article's field with a periodic potential.
The field. $a$ is real and periodic, $a\equiv a+2\pi f_a$; $f_a$ is the axion decay constant, with dimension of mass. The dimensionless angle is $\theta=a/f_a$.
The gauge sector. $F_{a'\mu\nu}$ are the colour field strengths, $g_s$ the strong coupling, and $\tilde F$ the dual; the density $F\tilde F=\sum_{\mu,\nu}F_{\mu\nu}F_{\mathrm{dual},\mu\nu}$ is a total derivative.
The potential. $V(a)=\Lambda^4\bigl(1-\cos(a/f_a+\bar\theta)\bigr)$ at leading order, with $\Lambda^4$ the topological susceptibility and $\bar\theta$ the vacuum angle.
The sign convention. The kinetic term enters with $+\tfrac12$ and the potential with $-$, so that the energy is positive and the minimum is the vacuum.
The Action in Element Form
The three pieces. Kinetic, potential and anomalous coupling,
$$ \mathcal{L}=\tfrac12\sum_\mu\partial_\mu a\,\partial_\mu a-\Lambda^4\Bigl(1-\cos\frac{a}{f_a}\Bigr)+\frac{a}{f_a}\frac{g_s^2}{32\pi^2}\sum_{\mu,\nu}F_{\mu\nu}F_{\mathrm{dual},\mu\nu} , $$
after absorbing $\bar\theta$ into the origin of $a$. Each piece has a different status: the kinetic term is exact and canonical, the potential is generated at low energy and is the leading term of a series, and the coupling is a dimension-five operator suppressed by $f_a$.
Why the coupling is dimension five. $a$ has dimension one, $F\tilde F$ has dimension four, and $f_a$ has dimension one, so the term has dimension five and is non-renormalizable by power counting. That is not a defect: the axion is an effective field, and $f_a$ is the scale of the physics that produced it.
Why the coupling can be written as a derivative coupling. Because $F\tilde F$ is a total derivative, $F\tilde F=\partial_\mu J_{\mathrm{CS},\mu}$ with $J_{\mathrm{CS}}$ the Chern–Simons current, the interaction is equivalent, up to a boundary term, to
$$ \mathcal{L}_{aGG}=\frac{1}{f_a}\sum_\mu\partial_\mu a\,J_{\mathrm{CS},\mu} , $$
which is a derivative coupling and therefore has the same soft behaviour as a Goldstone's. The equivalence is exact on a closed manifold and fails only when the boundary term matters, which is the instanton's contribution.
Why the potential is a cosine and not a polynomial. The periodicity $a\equiv a+2\pi f_a$ excludes every power of $a$ except the constant, so the potential must be a function of the angle. A cosine is the leading harmonic; the exact potential is a Fourier series whose coefficients are the correlators of the topological charge at each winding number.
The pairing. The pairing is the real scalar part, and the axion being a pseudoscalar does not change it. The field is real and the action is a real functional; the pseudoscalar character is a statement about the field's behaviour under parity and not about the action's reality. The coupling $aF\tilde F$ is a pseudoscalar density, so the integrand is a scalar and the action is real.
Why the coupling's sign is physical. $F\tilde F$ changes sign under parity, and $a$ changes sign too, so the product is invariant. The relative sign between the coupling and the vacuum angle $\bar\theta$ is what determines whether the minimum is at zero, and it is fixed by the requirement that the CP-conserving point be the vacuum.
The type of the coefficient. Three dimensionful numbers with three different origins. $f_a$ is the breaking scale and is the coupling's suppression; $\Lambda^4$ is the topological susceptibility and is the potential's height; $\bar\theta$ is the vacuum angle and is a parameter of the gauge sector. The axion's mass is the combination $\Lambda^2/f_a$.
Why the mass and the coupling are tied. Both are controlled by $f_a$: the coupling is $1/f_a$ and the mass is $\Lambda^2/f_a$. The tie is the reason a light axion is also weakly coupled, and it is what makes the axion a viable dark-matter candidate in the window the corpus records.
The selection rule. The selection rule is the shift symmetry plus the anomaly, and the combination is exact. The classical action is invariant under the shift; the measure is not, because the fermions that carry the Peccei–Quinn charge are chiral; and the anomaly's contribution is the $F\tilde F$ term. The action written above is the anomaly-included version, and the selection rule that produces it is the symmetry broken only by the anomaly.
What reality does not buy. Reality does not buy a mass, a scale, or a solution to the strong CP problem. Three things the reality of the action does not supply:
- No mass without the gauge sector. The potential comes from the same fields the axion couples to; without them the axion is massless.
- No fixed $f_a$. The decay constant is a parameter, and the allowed window is phenomenological.
- No automatic solution. The mechanism sets $\bar\theta_{\mathrm{eff}}=0$ only if the potential's minimum is there, which requires the CP symmetry of the gauge sector; the argument is the parent's.
The Variation: The Wave Equation with a Topological Source
The variation. With $\delta a$ vanishing at the boundary,
$$ \delta S=\int d^4r\;\delta a\Bigl[-\tilde\nabla_\Phi^{\natural}\tilde\nabla_\Phi a-\frac{\partial V}{\partial a}+\frac{g_s^2}{32\pi^2f_a}F\tilde F\Bigr]=0 , $$
so that
$$ \boxed{\;\Box_\Phi a+\frac{\partial V}{\partial a}=\frac{g_s^2}{32\pi^2f_a}F\tilde F\;} $$
which is the parent's equation. The sign of the source is the one that makes a positive topological charge push the field toward positive $a$, which is the convention that makes the CP-conserving minimum sit at $a=0$.
Why the source does not affect the minimum at leading order. The topological density integrates to an integer and is a total derivative, so in a constant-field background it vanishes. The potential's minimum is therefore at $a/f_a=-\bar\theta$ up to the instanton's contribution, and the instanton's contribution is what shifts it to zero.
The mass, read from the action. Expanding the potential about its minimum gives $m_a^2=\Lambda^4/f_a^2$, which is the second variation and not a separate parameter. A large $f_a$ gives a light axion, and that is the mechanism's whole point: the mass is suppressed by the scale that suppresses the coupling.
The shift, checked in the action. Under $a\mapsto a+2\pi f_a$ the kinetic and potential terms are invariant, and the interaction changes by
$$ \delta S=\frac{2\pi f_a}{f_a}\frac{g_s^2}{32\pi^2}\int d^4r\,F\tilde F=2\pi\cdot\frac{g_s^2}{32\pi^2}\int F\tilde F=2\pi\varrho , $$
with $\varrho\in\mathbb{Z}$ the topological charge. The phase $e^{i\delta S}=1$, so the shift is a symmetry of the quantum theory, and the periodicity is consistent.
The Constraints the Action Imposes: Periodicity Is a Constraint on the Potential
The constraint. The field is an angle, so the potential must be periodic with period $2\pi f_a$. That is a constraint on the functional form and not on the variation: the axion is unconstrained in the Dirac sense, and the periodicity restricts which potentials the action may carry.
Why the constraint is not a gauge symmetry. A gauge symmetry identifies field configurations; the periodicity identifies values of the field, and it is a statement about the target's topology. The axion's target is a circle, and the circle's radius is $f_a$.
The consequence for the potential's form. A polynomial potential is excluded; a cosine is allowed; the exact potential is the Fourier series of the topological susceptibility. The constraint is what makes the axion's potential a periodic function and therefore what makes the CP-conserving minimum one of a family rather than the unique minimum of a parabola.
The consequence for the coupling's form. The coupling must be a total derivative or a periodic function of $a$, or the shift would not be a symmetry. $aF\tilde F$ is both, because $F\tilde F$ is a total derivative; a coupling $a^2F\tilde F$ would not be periodic and would break the shift.
The Structural Consequence: A Potential Generated by the Same Fields That Break the Shift
The statement. The axion's potential and its anomaly are the same object read at different orders: the coupling $aF\tilde F$ produces, at second order in the gauge fields, the potential $\Lambda^4\cos(a/f_a)$ whose coefficient is the topological susceptibility.
Why this is structural and not a coincidence. The potential is the vacuum energy's dependence on the angle, and the angle is the axion's value; the dependence is generated by the instantons of the gauge sector, and the instantons are the configurations that carry the topological charge the coupling multiplies. The two facts are one fact.
The consequence for the mass. The mass is $\Lambda^2/f_a$, so the axion is light when $f_a$ is large and the susceptibility is small. The hierarchy between the axion's mass and the weak scale is the hierarchy between $f_a$ and $\Lambda$, and it is a consequence of the action's structure.
What the structure does not give. It does not give $f_a$, it does not give $\Lambda^4$, and it does not prove that the axion exists. It gives the form of the theory for any values of those numbers.
The Noether Currents: The Peccei–Quinn Current and Its Anomaly
The current. The shift symmetry's Noether current is
$$ j_{\mathrm{PQ},\mu}=f_a\,\partial_\mu a , $$
and its divergence is the anomaly,
$$ \sum_\mu\epsilon_\mu\frac{\partial j_{\mathrm{PQ},\mu}}{\partial R_\mu}=\frac{g_s^2}{32\pi^2}\sum_{\mu,\nu}F_{\mu\nu}F_{\mathrm{dual},\mu\nu} , $$
which is the statement that the current is conserved classically and not quantum mechanically. The divergence is exactly the source in the equation of motion, and that identity is the reason the axion's potential exists.
Why the anomaly is not a defect. An anomalous current is one whose conservation fails at one loop, and the failure is the term that generates the potential. The axion's mechanism requires the anomaly; a non-anomalous Peccei–Quinn symmetry would give a massless axion and no solution to the strong CP problem.
The topological current. The Chern–Simons current $J_{\mathrm{CS},\mu}$ is conserved identically, and the axion's coupling is its divergence contracted with $a$. The two currents' divergences are the same object, and that is the precise sense in which the anomaly and the coupling are one term.
Energy–momentum. Translation invariance gives $T_{\mu\nu}$ with the scalar and gauge contributions; the trace is nonzero because of the potential, and the potential's contribution is the vacuum energy.
What there is not. No second independent current; the axion's current is the shift's, and its anomaly is the gauge sector's.
What the Principle Adds and What It Relabels
Adds.
- The consistency of the shift as a computation. The periodicity is checked by evaluating the action's change under the shift, and the integer topological charge is what makes the check pass.
- The mass's origin. $m_a^2=\Lambda^4/f_a^2$ is the second variation of a potential that the action contains, not a separate input.
- The equivalence of the two couplings. The $aF\tilde F$ form and the derivative-coupling form are the same term, and the action shows it by integration by parts.
Relabels.
- The "strong CP problem". It is not a fine-tuning of a parameter; it is a statement that the vacuum angle is an observable, and the axion promotes it to a field.
- The axion's lightness. It is not a tuned small mass; it is $\Lambda^2/f_a$ with $f_a$ large.
- The potential. It is not postulated; it is generated by the same instantons the coupling counts.
The distinction that matters. The action has a shift symmetry and an anomaly. The potential is what the two produce together, and every property of the axion follows from that production.
Open Questions
Does the corpus's periodic central field have a unique origin? The parent constructs it as the phase of a complex central element; whether that construction is forced, or is one of several, is open.
Summary
The action. Kinetic term, periodic potential, and an anomalous coupling $aF\tilde F/f_a$ that is a total derivative. The variation gives $\Box_\Phi a+V'(a)=g_s^2F\tilde F/(32\pi^2f_a)$.
The three findings. The periodicity is consistent with the anomaly because the topological charge is an integer, so the shift changes the action by $2\pi\mathbb{Z}$. The potential must be periodic, which is why it is a cosine and not a parabola. The mass $m_a=\Lambda^2/f_a$ and the coupling $1/f_a$ are tied, which is why a light axion is weakly coupled.
The boundary. The periodic central field is the algebra's; $f_a$, $\Lambda^4$ and the existence of the axion are not.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $a$ | axion field, real and periodic |
| $f_a$ | axion decay constant |
| $\theta=a/f_a$ | the dimensionless angle |
| $\Lambda^4$ | topological susceptibility, the potential's height |
| $\bar\theta$ | vacuum angle |
| $j_{\mathrm{PQ},\mu}=f_a\partial_\mu a$ | Peccei–Quinn current |
| $J_{\mathrm{CS},\mu}$ | Chern–Simons current |
Further Reading
- The Axion and the Peccei–Quinn Mechanism in Biquaternionic Form — the parent: the mechanism, the potential, the window.
- The Lagrangian and the Goldstone Effective Action — the derivative-only action whose shift the anomaly breaks.
- The Lagrangian and the Central Scalar Field — the field the axion is, before the periodicity.
- The Theta Vacuum in Biquaternionic Form — the vacuum energy the potential promotes.
- The Theta Parameter and Strong CP in Biquaternionic Form — the problem the axion answers.